Medical Image Understanding and Analysis took place in Leeds on July 6th and 7th. There were 116 delegates (of whom 34 were BMVA members), 28 speakers (representing 20 institutions) and 12 posters. Voxar and Floating Point Systems exhibited. So much for the statistics.
MIUA (only pronounceable as "miaow") was first held in Oxford in 1997 and provides a forum for the whole range of medical imaging interest - this is very broad. This breadth was accommodated successfully by running a single track conference in 7 sessions giving space to all topical issues; a simple calculation reveals that individual presentations were short (15 minutes plus questions) and the session chairs were careful to keep things on schedule. Such a constraint on time turned out to be very good discipline for the speakers, and talks were characterised by high quality delivery and AV in which speakers of necessity went straight to the point, with no opportunity to use bandwidth on tangential issues. Printed proceedings provided the references and supporting material the audience might have required.
What any individual might regard as the highlight of the conference would depend on personal interest and it might be hard to find a majority view. Everyone this reporter spoke with over coffee agreed that there was "something for everyone", and it would be inappropriate to single out any particular topic for special mention. Having said that, two themes which surfaced more than once that took personal attention were evaluation, and the nature of demand for 3D data. Evaluation, both of machine systems against clinical opinion and of various clinical opinions among themselves, provided some interesting and provocative results. Whilst it may not be true to say that machine systems are trying to hit a moving target, the target's precise location may depend on who you ask. Similarly provocative was a clinician's enquiry about whether 3D information was actually necessary; all those designing visualization systems, please note.
The conference was dominated by computer scientists, mathematicians and physicists; indeed, on the first day a question to the audience elicited the information that only two people in the room had medical qualifications - this was an underestimate as various clinicians made themselves known over the two days, but it remains true that this was a "computer" event, augmented by a respectable contingent of mathematicians. Many of the talks were given by PhD students and this provided an excellent opportunity for them to gain experience in front of the expert community; they all acquited themselves very honourably.
The posters provided content as interesting and varied as the talks, but of more variable quality. Poster authors sometimes need to remember that this is a mode of presentation that requires more than verbatim translation from a paper or prepared talk.
The Conference does not award prizes, but a personal view was that for quality of presentation and delivery, Alan Jackson's talk "Improving time of arrival map quality in MR perfusion" stood out, while the best crafted poster was produced by Shark et al. of Central Lancashire on virus classification. There were many quotable quotes: The "If it works, leave it alone" award goes to Neil Thacker (Manchester) for his remark "I don't care about Fourier domain characteristics"; the "Best answer to an illustrious professor of statistics" award goes to Graeme Penney (Guy's) for "I'm just going to have to nod"; the "Be careful with this" award goes to Hava Lester (UCL) who counselled care with her warping procedure: "It can be dangerous if you don't know what you're doing - especially if you're doing brains"; and the "Most applicable work" award goes to Daniel Poxton (Manchester) for the observation "Neuro-scientists assure me this is of some kind of interest". The "Making academics feel uncomfortable" award was won easily by Paul Taylor (UCL) for his remark describing "a whole bundle of measures that work well enough to get published but not well enough to get used". Mike Smith's (Leeds) wildly inaccurate introductory remarks describing Leeds architecture are best left unrepeated.
Congratulations go to Liz Berry of Medical Physics in Leeds who managed a thoroughly interesting conference, complete with highly successful dinner; the only technical hitches were attributable to the builders next door who began drilling to Australia (but were quickly halted) and to Bill Gates, whose software performed as normal (Liz was not up to solving this one ...). Details of MIUA, including information on next year (Oxford) and availability of this year's proceedings are at http://www.miua.org.uk.